1999 - Prediction Of Hospitalization By Schizophrenia Patients' Assessment Of Treatment - An Expanded Study

JoueNaror Rvcsrarrlc PERGAMON

RESEARcH Jouhal of ?sychiatncResa.cn ll (1999)llfl

I9

Predictionof hospitalizationsby schizophreniapatients' assessment of treatment:an exoandedstudv StefanPdebe*,Matthias Brdker D.parlhlert ol Sortol Psf.hidir,Ft.ip

Uhiterliit

Be.lin, Platanenol?e 19,l.n5A B.rli..G.rnanr

Rc*1vcd 20 Nov.mber 1997;receivedn revised fom?July

l998i.cepred

l0July

1998

patientsglobalassesmen!of lreatmenrpredictsoutcomein communitycaE. Ei8hty, This stxdyexamines whelherschizophrenia five patientsrated the extenl to which their tre.tment was riSht lor them. The oulcomemeasurewas assssedby meansof a hospitalization indexrefleclingthe durationof lull and partialhospitalizadon wirhin a follow,upperiodof24 ironlhs. Patienlswho droppedour(r:21)assessed theirreatmentmorenegalivelythanlhosewhoremainedin thecaresystemduringfollow-up.Palients (n : 25) cxpreseda more negatileaseshent than patientswho werehospitalizedin ihe follow-upperiod wrlb no bosprialization at least once {r : l9). ln this lalter subgroup,however.paiientswith a more negativ€assessnerlhad a loDgerduralion ol hospitaliraiiondunng follo*-Lrp, and palientJ asesmenl of lreahenl was the best singlepredictorof outcorne(r= -0.50, P<0.01) In a muhiple reeression ualysis, patients askssmentof lreamenr. the nunber of preuous hospitalizations and the hosfitalizatlonindex in the lwo tears prior 10 the inteniew, together,explainedapproximatelyhalf of the varianceof ourcome (adjustedR: = 0,{6) The nndingsunderlinethe lelevanceof askin8patienrswherherteatment is rjgh! for theln. allhough the relatiorshipberweenpalienrs'assessmenl ofbeatmentand long-tem oulcomesems nore conplex thanprevioussludiessuggesled. a 1999Ehelier ScienceLtd. A11rishls reserled.

l. lntrodmtion In variousstudies,schizophrenia patients'assessment oftrealmenthasbecnfound to be a significanlpredictor for outcome.Patients$ho do not believethat thei. cur rentpsychiatrictreatmentis rightforrh€m,showa poorer compliancewith that treatmert and ar€ more likeiy to terminatetr€atmentearly. Even if rreatmentis applied conlinuouslyand properly,patientswith a morenegative assessmen! o{ lreatmenttend to have a lessfavorabl€ outcome.In shorr-lermneuroleptjctreatm€nt(Singh, 1976:Van PullenandMay. l978iVanPullenetal.,1981, 1984;Hoganet al., i985: Bartko et al. 1987iHogan and Awad, 1992;A$ad, 1993).in compl€xhospitalireatmenl (Briiker €t al , 1995).and in day hospital treatment (Pdebe,1992;Priebeand Gruyt€rs,1994)patientswho initially assessed treatmentmorenegaiively,show€dsignificantlv less improvemenlol psychopathologyafter threeor four weeksof medicationor at dischargefrom hospitalor day hospitalrespectively. ln all thesestudies.

'Cor€sponding author Departdemof Psycnolosical M€dici.€.St. Banholomcvs and The Rotal London School of Medicine.Wesl Smidrneld,London ECrA 7BE. UK. Tcl : + -Frr4/rlJtt586 52?1 fax: + +44rl8l'586-5211

patients' inilial assessment of treatmentwas obtained using simple questionssuch as to what extent they b€lievedthat their medicationor treatmenlin the given s€ting wasright for themWhile most studiesexaminedthe r€latjonshipbetween patients assessmentof treahent and short-lerm outcome.we investigatedin a pr€viousstudy whether suchassessments alsohavesomepredictivevaluelor the outcomeoI long{€rm treatment-Using a visual analog patienlsreceivinglong{erm treal scale.34schizophrenia m€nt in communitycareratedthe ert€nt to which their pretreatmentwas right for them. Thes€assessments dicted the duration of full and partial hospitalization during an initial follow-up period of 12 monihs and a longer follow-up period of 30 nonths. Pahentswho assessed theirtr€atmentmorenegativelyhad significantly Iongerhospitalizations in the two follow-upp€riods.Th€ predictivevalueof palients'ass€ssments oftreatmenlwas not explainedby the influenceofother variables(Prieb€ and Gruyters,1995).This wasa preiiminaryfindins in a small and somewhatselectivesample.If it is. however, replicatedin further studies,it may haveobviousimplicationsfor clinicalpracticeas well as for research. Two aspectsappearpartrcularlyrelevant:(a) This predictor can be obtainedby an extremelysjmplemelhod which can be applied under most circumstances and is inex-

0022 19561991$ seetronr mart( ie, 1999Flevr.rSoence Ltd Al ngh15rc*Ned. P I I :S 0 0 1 2 1 9 5 6 ( 9 8 ) 0 0 0 1 75

l14

S ln b.. M Brakt iJautwl.fPtt.hialri.Res.dr.h3311999 1 )1 3 1 1 9

pensive;(b) Patients'asscssmcn! of treatmentseemsto be a predictorin its own right and may explarnvanance i n o u r c o mi en r d d i r i o nr oo r h e n r r e d i c t o rl.h. r ' a ( u m p tion is in line with resultsfrom researchon consumer satisfaclioniallhough patientswith a higher degreeof psychopathology lend 1obe lesssalisliedwilh trealment inmosl studiesexaminingthisrelationship.and although correlationsof consumersatisfactionwith other sociodemographicor clinical variableshav€ been reported inconsistenily. the associations areweakso that parients' satisfactionwith rreatmentor assessment of it can not simply be taken as an epiphenomenon of other factors (Dennerand Halprin. 1974;Edwards et al., 1978;Larsen et al.. 1979tLevoiset a1.,l98l; Berger.l98l; Hansson and Berglund.1985:Hanssonet al.. 1987;Conreet al.. lq8o:Mrlnt\reeLul . lq89:Crurte|.andPriebe.lrlr))). Sincetheoriginalstudy.wehaveexamineda bisgerand more compfehensivcsamplcof schizophreniapatierts receivinglong-rermlreatmentin community care. We investigated whethertheir globalassessmeni of treatment predictedthe outcomewithin a follow-up period of 24 monlhs.Threequestionswereaddrcsscd: (l) Do palienls\rho dropoutol lrcatmenrin commudty careduring rhe foliowing two years,expressa more negativeassessment of treatment? (2) Doespatients'globalassessment oftreatmentpredict ihe duration of their full and/or parrial hospiralr.,"run\ rrhrarhetollow-upne'rodIn rhe cmarnrng groupr (l) Is the predictivepowerof theseassessments €xplained by the influenceof other variablesor do€sit add to thc predictivevalucofothcr factors?

2. Methods This studJrwas carriedout in the samesetlingas the o r i p : n ai ul d y I r i , r t o . n m n i r \ c r r e . ) i e m i i a n i q n ( ' dlstrict of Berlin.Germany(Charlottenburgwith aboul 200.000inhabitants)(Steinhartand Priebe. 1992).The syslemis orientedtoward providinglong{erm tr€atmenl for palientswith severeand chronicmentaldisorders.h includesrhrecpartial hospitalizalionprograms.several communi!],basedservices(a day-carecenter.a drop-in center.singleand group protectedliving apartments). a n d v a f l o u o, u r p - r r e nf irc i h r i e ,I.n - p a | | e cn ar ' eh p ' o vid€d in co-operaiion\:rith psychiatdchospitals.Continuily of care is guaranleedby clinical casemanagers (Priebe.indCruytcrs. I 995). Crirenafor inclusionwerea diasnosisofschizophrenia or schizoaflectivedisorder according to DSM-lll-R (Americdn P.)chidrncA..ociarion. lr)8 )andconln ou, ireatmentin the communitycaresyslemfor at leastthree monlhs.All 120palientswho fulfilledthesecriteriawirhin a period ofone year.wercincluded.Patientswereasked

to assess their treatment("Is the treatmentyou are currently receivingright for you?").Th€ answerswereselfpoint': raledon a l0cm longvr.ualanalog'cale {e\rreme 0 = not right at all. l0 : completelyright). Each I cm intervalis markedso that the scalecombinesqualitiesof a \ r ' u J ja n a l o g s L d l \er r r hf e a r u r eosf d l l - p . ' n r r a r r n g scale (Luria. 1975: Guyatt el al.. I987; Priebe and Gruyters.I 993.I 995).The samescalehasalsobeendemonstratedto predictoutcomeof day hospitaltreatment and of in-pa||enrrrearmenr In ,chi/ophreniaparienrs (Pri€beand Gluyters, 1994;Bf.tker et al., 1995).Socioderno-qraphic vaiables and data from patients psychiatdchistorywererecorderlusjnga standardized documentalion (Bosch and LUbke-Westermann.l98l). Psychopathologicai symptomswere rated on the Brief PsychiatricRating Scale(BPRS;Overall and Gorham, 1962).All pati€ntswereexaminedby thesameintervrewer not otheMiseinvolvedin treatmentwho also madethe diagnosison the basisof a PresentStateExamination (Winget al..1974). Th€ total numb€rofdays of both full and partialhospitalizationwas recordedfor the follow-up period. A hospitalizationindex (HI) was calculatedrefleclingthe desre€(full or partial) and duration of all periods of hospitalization:(x days in full hospitalizationx 3)+ (" days in partiat hospitalizationx 2)i(daysin the observation period).The HI a modifica.ionof indicesused by Lavik (1983)and Tansellaet al. (l986Fprovides an approximalerelleclionof costs fbr in-palienl and day hospitalcareduringthe observationperiodand hasb€en used for assessing long-rermoutcomein other studres (Steinhartand Priebe,1992;Pri€beand Gruyt€rs,1993, lqq5).TheHI ua. al.ocalculared for lhe t$o yea|.prior The methodsfor €xaminingpatients' assessm€nt 01 treatment.psychopathology and outcomewerethe same asin the originalsludy.Because therewasno substantial dilierencebelweenthe resultsconcerningthe lwo followup periodso[ 12 and ol-30monthsin the original study. we usedonly one foliow-up period in this study.Thus, there was only one outcomecriterion and one hypothesizedpredictorsothat Bonferroni-adjushentsofpr€dictiver€lationships werenot needed. 2.1. Statisticalanalrsis Differencesbetween subgroups.i.e. drop-outs vs remainingpatientsand hospilalizedvs non-hospilalized patients,wereanalysedby meansofx'contingencytables ard l-testsdep€ndingon wherherthe variablesinvolved were conlinuousor cateeorical.when ditrer€nces wer€ signlficant,group difi'erences were controll€d for th€ influenceof othervariabl€sby a simplefactorialANOVA inciudingothervariabl€sasco-variates. Thus,ditrerenc€s were reanalysedwith the influenc€of other potendally For eramininSa"ociinfluenlial\ariable'parlidlleJout

\ P t e r ? u a t . I p , J u a t t . I P . , , h i n t , nR . , n t , r , ,

of tr€atmentwith sociations of patients'assessmenl odemographicand clinical vanablespoint-biserialcorrelations with dichotomousvariables,and Pearcon's producrmomentcorrela!ion. $ilh conrinuous\afithe contributior of variableswer€ used.For assessing ablesto the predictionof group differences we applieda stepwisefo ard logisticregressionanalysiswith group sialusas dependentvariable. The way in which patients'assessment of treatmenl and orher variabl€spredictedthe HI in the follow-up periodwasexaminedby singlecorrelations. Additionally. analysiswith the a siepwiseforward multiple regression Hl as dependentvaiable was calculatedincluding all vadablesshowingsignificantsinglecorrelationswith the HI as prediclors.Accordingto the centralUmit theory ' of rhemulrrflereCre(rron $e co1'rdered lherequrrmenr analysisregardingthe distributionof valuesas fullilled in a samplesizeof >10 (Kendalland Sluart, 1973). 3. Results 3.1. Sanple

laao ltt tla

medication.The meanneurolepticdosagewasequivalent to 340mg (SD : 270)chlorpronazine(CPZ) (Jahnand Mussgay,1989).57 patientsregularlyparticipaiedin one of $e following group actlvities:occupationaltherapy. musictherapy,dancing.larious forms of talk therapies. accommodation. and 24in Ten patientswerein sheltered jntensiveday careprogrammes. 3.2. Atsesflent ol trcatmenl On average.paiients'vicwol long-termtreatmentwas positive(mea. : 7.4. SD : 2.4). The ertent to which p"l eal' raredrher rreolmenr"' berneriehL$as .iEnilicantly correlated with ase (Pearson's | : 0.23, P < 0.05),to the BPRSscore(r 0.23,P < 0.05).and to the HI during lhe two yearsprior to the interview (/ : 0.28.P < 0.01).Youngerpatients.palientswllha higher degreeof symplomsand thosewnh longerduralion of hospitalizalionin the lasr i$o y€arstendedto assesstheir keatment more negatively.Other socio!ariables(gendef,livingsituationandoccud€mographic pationalstatus)or clinicalvariables(numberof previous hospitalizations, duration of iliness,and curent dosage medication) rlere not significantlycorof neuroleplic patients global relaledwith assessmenl oftreatment.

Thirty-fivepalienls did nor participatein the study; 25 of them r€fusedto be examined,and l0 were not inierviewedbecauseof the sev€rityof psychopathology. 3.3. Di|fercnrcshet\teensubstaqs These35 patientsachieveda lower levelof schooleduBy the €nd of the follow-up,2l patientshad dropped cation(x/ : 10.63,df : 3, P < 0.05)andofoccupational ()1-lhese qualification(x':9.41. dt:4. P:0.05) than the 2l patientsand our oftreatment.Characteristics remaining85 patients.The two groups dld nor djffer of the remaining64 patientsare shownin Table L The 2l patientswho la1erdroppedoui. assess€d their signiJicandyregardingany oth€r sociodemograpbic or lreatmentsignilicantlymore negativelythan rhe 64 patclinicalvariable. The agesof the 85 patientswho wereinterviewed(38 ientswho did noi. They also had a higherHI in lhe last women and 47 men) ranged from 2l {8 years two yearsand were more often unempioyed.yet these (mean:38.8. standarddeviation(SD:9.7)). At the There differencesfailed 1o reach stalisticalsignificance. was no significant ditrer€nce in any other socitime ofthe study.48 ofthem wereliving alone.21 with a padner.and l0 parrenLs $ere Ii!ing $ilh theirparents odemographic or clinicalvariabl€includingthe recorded or with other family members.Elev€npatientshad not tr€atmenicharacteristics. In a stepwiseforward logistic regression analysis,we lestedwhich variableswould sigcompl€tedtheir primary schooleducation,52 had comparienrs pletedsecondaryschool,and 22 had compleledhigher nrll(!nrl)conrribure ro rhepredrcr^nofwherher education.Tw€nty-ninepatientshad no occupational droppedout ornot. In thismultivariateanalysis.parients' qualiJications. 52 had completedan apprenticeship, and assessment oftreatmenttumed out io be the only predicTwentypatientshad a job; four held universitydegrees. tor. Thus,patientswho drop out ofireatmentduling the 65 wereunemployedor prematurelyretired.At th€ time r$o-)ear-follo$-up eipre..eJa more rcguri\eJ"e'of the inrerview the mean BPRS scor€ was 32.6 ment oI treatment,and this predictile .elarionshipwas (range: l8 55;SD : 8.5).Thedurationofillnessvared not €xplainedby the influenc€ofany othervariablethai betw€€n0and 39 years(mean: 10.9,SD : 7.6),and the wasrecordedin the study. numberof previoushospitalizationsrangedfrom 0 17 The remainingtwo subgroups\lho did not drop out (mean= 5.2.SD : 4.0).Patientshadbeentreatedwithin within rhetwo yearfollow-up,ditreredin somevariables ih€ communitycare systemfor betweentlree monrhs as summariz€din Table2. and approximately 20 years (mean: 5.8 years. As conparedto the 39 parieniswho werehospilalized :4.7). years prior pati€ntsiendedto SD The HI for the two to the study at leastonce.the 25 non-hospitalized $ a r 0 . 7 0( S D_ 0 . 8 1 )A r r h e r r m e o f r h e r n r e r . r e u . s e ' e n longer had been treatedin the havea duration ofillness. patientswerenot takjng neurolepticmedication:57 palcare system for a long€r time already, had a communily jents were receivingconventionalneuroleptics,and 2l lower HI in the two yearsprior to the study;and tended patientswere on atypicals;3l patientsreceiveddepot to assess their treatmentmorenegatively;when the jnflu-

S. Prieb.,M Bti'kzt i Joumalaf Psrthiut/n Ret?anhJi 1t999)t13 119

Sociodmographi. and cliniel charocIeflsd.s: dlfferenes berwen drop.uts &d prtienls .emaining in the care systen during ioilow-up

i0(4?.6) li(52.4)

23{41.3) 16i56.2) l9 219.?)

ll(61.9i

I t (41{,1) ll(51.6)

8(38r) S... eh@l nor conpletcd Se *hool ompkt€d

4(r9.0) I1152.4) 6128.6)

Prof6nonal quali,i€tion No @upalional qnalii€Iion AppF.tieship @nplet€d

DuBtion of 'lln€ss(t$rt Nmb.! of pEvious hospilaLaliois Tmhnt in ine e.€ sy$en {Foirht HI for the ldt $o y€ds N€uroleptics (CPZ equivalenls, '. ng) As$m6r orr.€atmcnl(on vAs)

8(18.1) l ] ( 5 24 ) 2(9.t 2(9.51 r9(90.t 9.4(70) 1.6(.21) s2.l{sr.8) 1.0{'.0) 15,r.,1i1374) 6.1(r.7) 137 (9 rJ)

4r(64.r ) r6(25.0)

0.t

I

1.2

2l (12.8)

4r(641) 20.t)

l

t8(281) I

l r.r(7r) s..1(4.11 14.5i56.4) 0.6(0 7) o) 334.5(24r

1.0

I 8

7 9 ( 2l ) l2.l(8.01

8l 83 8i 83

'Th€ di|l€E.e renainedsigniiort !fter cortrollire for effels ol lhe HI dnnnCla$ r{o y€rs. and ofdejob situdtion.

enceof HI during the lasrtwo years,duration of iliness aDli duration of treaiment in the care syst€mwere con rrolled(as co!driar$ in A\OVA). 'he diferencein parients' assessmentof treatm€ni reached statistical sieni{ic$cc. Includingthesefour vadablesas predictorsin a stepwiselogistic reg.essionanalysis the HI in lhe past of treatmentpredicted two yearsand patients'assessment signincantlywh€therpatientswere hospiBlizedduring the follow-upor not (r' : 18.2,P < 0.001). 3.4. Ptedicion of hospitalization The 39 patientswho were hospitalizedat leastonce during the two years following the interview, spent p'o hfl\rceno and
oi ireatmentwas more than a more negatrveassessment threerimeshigh€rthan the HI in the othcr 22 patients who had assessed lreatmentmorepositively(0.87 vs0.:7. l:3.7,P<0.001). Besidespatients assessmert of treatmenl.four of all the variablesrecordedin the study w€rc signilicanlly relaled1o the Hl in the subsequenttwo yearsrn lhese l9 patients.Patientswith a longcr duration ol illness (r: 0.33,P < 0.05),with moreprevioushospitalizations (/:0.41. P < 0.01).and with an higherHl in the last two years(/:0.43, P < 0.01).and thosepatientswho w€re unemployedor premarur€iyretired(poinl biscrial / :0.17, P < 0.05)spentsignificanrlymorc days in full andror partial hospitalizationduring the lbllow'up period-However,no variablerecordedin this srud-\r was patwith both rhe found to be significantlycorelated ients' assessment of trcatmentand the Hl. A stepwise multiple regressionanalysiswith thc HI as forward variable $as calculatedincludingtheseibur dep€ndent ol lreatm€nt vanablesand patients'globd assessnenis as predictors(i.e. all variablesshowingsignificanisingle conelationswith the HI werc lestedas prediclors).The resultis summarizedin Table 3. Patients'assessm€nt o{ treatmenrwas the b€st smgle

S Prieh?.M. Btijkt

Jaumalol Ps!.hbnit R.seat.h3J 11999)tti

tt9

and hospiializ.dpatients socrodenographi.andclinicalcharacteristics:diferencesbeNeennon_hospitalized Non hospinli4d pat8nis(,:251 Mean(SD)1n(%)

Sec.shool oo! conplcted Sec {hoolcohpleted Higbereducaxon qual lcaion Professional No occupaionalqnalifrcaxon Apprenliashipcompleted Unrve6rtydegree

Duradonol rllne$ (yeart Nunbcr of previoushospitalizllron: Trcatnenrin the caresysteF(months) HI for the last twoyears Neuroleplics(CPZequlvalens.in ms) A$e$nenr of treahe.t (o! vAsl

Hospitahed palients(n = 19) Mean(sD)r,(%)

.11'

df

1t (4,1.0) 1 4 ( 50 6) 4 t . 4 ( lr 4 )

l7(.11.6) 22(56.4) l7 8 (lJ3)

00 1.4

r 62

l 2 ( 4 80 ) l 3 ( 5 20 )

r9(48.7) 20(5r.3)

0.0

r

r4(s6.0i I (zrJo)

1\1.7) 27t.69.2) 9(21r)

1.5

2

3.4

2

1.3 19 06 20 :t.6 0.1 l 8

l 62 62 62 6: 62 62

r.0

62

?(28.0) l6(64.0) 2 ( 30 )

r4(35.9) 25(6,1.1)

9060)

9(21.r)

l r . 5( 8 . 0 5.8(,1.6) 9 l r i ( 5 24 ) l3rJ.2(2ll.l) 7.1(1.4) ll I (?.5)

99(7.2)

s l(1.r) 63s 66.6) 0.8(0.ri) ll2 2(212.31 8.1(1.9) l l . 0 ( 84 )

P

<0.10 <0.05 <0.01 <0.1e

"The difierenLtrelcheda high levelol lgniticar€(P < 0 0l) after cont.olb.g lor edmrsof Hl dunns lasi two y.ars, duralion oi illne$, and duralon of treatnenrin the cdresyst.n.

predjctor of outcome.The frequencyof pr€vioushospitalizationsand the HI for the two y€arspnor to the srudy w€re also includedin the equationof regression and contriburedsignificantlyto the explanationof the variance.Togelher,rherhr€€variablesexplained5l% of the varianceof lhe Ht in the follow up period.

4. Discussion patients'globalassessment ot treatment Schizopbrenia in communitycarewas,on average,very posilive.Yet' il vaied, and its variationwascorrelatedwith patients'age with Thesefindingsare consisteDr and psychopathology.

index(HI) for 24 sonths (, - l9): signincantcotelalions Predictionoffte hospitalization wlth prcdidon and tepwis lo$lrd Fultifle regresion analysn (Mullillc R-o.?l Rr = 0 51.adjunedx: = 0.46)

iollow-np

A$e$nent .f tFaheni Prevr.ushospiBlisatio.s Hlduring lasttso yeaN O c d p a r o n a \l u u .

0.501t 0 41" 041** 0ll* 0l7l

MultitleEgnsionaDalysis

I

^rchanee

-0.37*' 043r' 0l5'

S l e pl : + 0 . 2 5 S t e p2 : + 0 . 1 5 l:+ 0.11 Step

tP < 0 05r1* P < 0.01;ns not signrlicantinor in lhe 6nal equalion.

ll3

S. Priebe, M Brtik?r i Journal ol P srr hiani. Res.ur( h 33 | I 999) I I J I | 9

numerouspreviousresultsreportedin theliteralure(Corrigan, 1990;Gruytersand Priebe,1994). Padentswho droppedout of treahent duringth€ two yearfollow-up,had expressed a significantlymore negative assessment of treatmentin the interview.lt is not surpnsingthat patientswho havemores€riousdoubtsas to whethertheir treatmentis right for them, are more likelyto terminatethistreatmenlearlythanthos€patients who are convincedthat their current treatm€ntrs compl€telyright for them. This resultmay be seenas a predictive validity for lhe simplemethod usedto examin€ paiients assessment of treatmentin this study.Despite the lrivial nature of this result it may nevertheless b€ relevant.Patients'assessment of treatm€nt\ras the best predictorfor whetberpatientslater droppedour or not, and in a multivariateanalysisits predictivepower was not improvedupon by the additionalconsideration ofall othervariablesrccordedin this study. Wilhin th€ subgroupofpatientswho did no! drop out and who werehospitalizedat leastonc€in the follow-up period,patients'globalassessment of treatmentwas th€ bestpredictorfor the durationoffuil and/orpanial hospitalizationdDring the following two years.Alone, it predicted25% ofthe varianceofthe HI. This predictiv€ value was not explainedby the influenceof ary orh€r variablerecordedin this study.Togetherwith threevariablesof patients'psychiatrichistorypatients'assessment of treatmentadd€dto an explanationof approximntely halfofthe varianceof theoutcom€criterion.This amount of predictedvariancemay not be satisfactoryand cerprediction lainly not good enough10makea reasonable lbra singlegivenpatient.but it seemssufficientto regard pati€nts'assessment oftreatmenlas a relevantpredictor. Theseresultsarein linewith tlle originalstudyandwith findingson the relationshipbetweenpalients'assessment and ourcome of short-term treatment. This study, however,is not only a replication,but an expandedinves, tigation.More variablesofpotentiai influencewereconsideredthan in the original study. and the samplewas less selectiveand bigger so that resultsfor subgroups could be analyseds€parat€ly. Th€ analysisofsubgroups revealedthat the relationshipbetweenpati€nts'assessment oflong-term treatmentand outcomemay be more complexthan previouslysuggested. Patientswithoui ary hospitalizationduring follow-up had exprcsseda less positiveass€ssment oftreatmentthan tbosepatientswho werehospitalizedar leastonce and not a more positive one as one might haveexpected.The differcncereached statisticalsignificance when the influenc€of oth€r influ€ntiallaciorswascontrclied. The subgroupof patientswith hospitalizationduring follow-uphad assessed theirtreatmentwithameanscore of 8.2 on a scalethe positiveextremeof which is 10. A more posiliveassessment on a group level is hardly possible.Thus. there is an obvious limitation to predicting a more favorableoutcomeby a more positive

assessment of lreatmentacrossgroups,In groupswilh psychiatric different historiesand prognoses,patients' assessment of treatmentmay adjust to differentlevels. Moreover,it might reflectdillerent psychologicalproin the patients.It can be specucesses and expectations patients latedthat who werenot hospitalizedin the two yearfollow-up,had higherexpectations as to what their treatmenrshouldachi€veandthat theyw€relessprepared to acceptdisadvantages and sideeffe€tsassociated with treatment.The courseof their illnesshad been more positivein the last two y€ars,and they might seeless need for psychiatrictr€atment.want to becomemore independentfrom it and subsequ€ntly feelmore critical toward their ongoing treatment.In this subgroup.the vaiation in patients assessmenl of treatmentcould not be t€stedlor its predictivevaluefor outcome,because the oulcomecriterion used in this study did not vary (all palienrshavinga HI ofzero). It remainsunclearwhetherthe findingscan be replicatedin dillerentsettingsand in patientswho are not in continuousireatment for, on average,more than five yearsalready. Furtherstudiesshould focuson the mediating factors accouniingfor the predictiveassociation pati€nts'assessment betweenschizophrenia and outcome of long{€rm treatment(Priebeand Gruyters,1993).It may be noted that concemingoutcomeof short-term treatm€nt those factors are also poorly understood. althouehnore research hasbeendoneinihat freld(Singh, l9T6iVanPuttenandMay, 1978;VanPuttenetal., 1981. 1984tHoganet al., 1985:Bartko et al.. 1987;Hogan and Awad. 1992iAwad. 1993).In long-termtreatmentthis predictiverelationshipis evenmorecomplicatedbecause the associationis as this study shows noi straightforward acrossdifferentgroups,because th€ assessment of treatmenthasbeenshownto vary substantiallyover longer periodsof time in somepatienls(Gruytersand Prieb€,1992),and because aimsof treatmentmight vary over time and acrossindividual patienismore than in ,horFrermrrearment. In lhis srud).hospilali/dIon\,tas usedasan operationalized outcom€criterion.Whil€ preventingand sho(eningof hospitalization is a cenlralgoal in community care. it is not the only one; moreover. reasonsfor hospitalisation aswellasits socialandclinical impactvary dependingon the givenlreatmentsrtuatron. Further r€searchmight use more specificand individualizedoutcom€criteriaand shouldexaminewhether specifictherap€uticinterventionscan aff€ct patients' assessment oftreatmentand.subs€qu€ntly, improveoutln line with previousstudies,the resultssuggestlhat patients'globalass€ssment schizophrenia is of predictive valuefor the outcomenot only ofshortt€rm tr€atmeni, but alsooflorg-term treatmentin communitycare.This applieseveniflhes€assessm€nts are obtain€dby a simple method.A more negativeassessment is associated with a higherlikelihood to drcp out of treatment,and wirh a

\ p,"L,,u

atar,,

J o w a o ,a l t , , , h t r t ,

lendencyfor longer petiodsof full andior partial hos' pitalizalionsin thosepatientswho do not drop out and are ho.prrdlledal led.ronce The findinglhal palienr. withour any hospiulizationin the follow-upperiod,had assessed their lreatmentmore n€galivelythan !h€ hosin paipitalizedpati€nts.demonstrates that differences i€nts' assessmenr betweensubgroupswith differentpsychiatric hisroriesand probably with ditrerentlevelsof expectations arenot fully und€rstoodyei.Tbus.palients' assessment oflong-termtreatmentalone.i.e.withoDtturther data specifyingthe patients prognosisand expeclations,hasonly a limited prediclivevalue. of Nevertheless,schizophreniapatients assessment psychiatriclong{erm treatment appears!o be worth obtaining,and one simplequestionis suliicientto elicit prediclive informatior. In clinical practice, patients' answersshouldbe takenseriously.Negativeassessments may be a reasonto €xplorepatienls'viewoflreatm€ntin a morederailedwayandto consid€rmodifyingtreatmeni componentsor the treaimentselting,In rcscarch.patrenl\' ralngs ma) be usedfor e\plarnrngvandnce,n preferabl)in addilionIo otherpredictors. ourcome.

Refercnces AmericanPsychialicA$ocDron DSNl.lll-R (Eds.) Diasnoslicand SratistralMdnull of MenralDisordes. 3rd rev edn \\rshrnslon: The A$oc1!t,on 1937. schizophreda. S.hizoAwadAC SubjectiveEsponselonenrolepticsi. phreniaBullednl99l:19:60916 Barlko c. HerczegI, Be*asyM. Predictingoutcone of neurolepl( lieatncnl on lhe bash of subjeclvensponseand eady clinical 5. inprolenent. Journalof CllnicalPstchiatlr 198?i48:361 B€rserM To*lrd naxinizing th€ utility of consunersalisfaclonar anoDtcone.ltr Lanber MJ, ChnsEnsenER. DeJulio SS.edito6. New York: Johnwiley TheAsFsnenl of PsychothcrapyOutcone. and Sons,193r.p 57 79. Bosh C, L0bke-WesFmannD Ein Syslemzur Dokunentationpal ientenbezogener Kontatle Der Nene.azt l98l:ll:283 l. lreahent BrdkerM,RohdchlF, Prieb.S.lnnialassessnentolhospital bt patientssnh paranoidshizophretuarA tredlclor ot ourcome. 3l PsycniaryResearch1995j58:77 conte HR, Plutchik R, Buctl€t P. warrcn spenceD, Byran Karasu liew their psychiatncrreatment.qosprlalandCon T. Ourparienrs nunity Pslchiaty 1989:.10r641 l. CorrisuPW Consunerernhctio. aith insritutionalandconnunit)_ care Conmunit! MentalHeallhJoumal1990j6:l5I 55. DennerB. Halprin F Clienlsand theraprsNelalute clinicalservices Am.ricanJournalof Connuniry Psychology1974i2l7l 8 D Doespatientsalisf&tion EdwardsD, YaRis R, M uellerD, Langsley Hospilal and Connunrll Psych,ari! codelare wilh iu.esl GruytersT, Prieb€S. Palienls viewof psychiakictreahent a study on pmblemsofits asesne.t and stabilityover tine (in Geman) 1,405 Fort$hrnte der Neurologie Psychiatrie1992160 of psychratnctrealment Gruyt€s T. PnebeS Uers assessmenr resulrsand problemsof a systenaticexdldaion (in Gem$). Psy.hiatn.he Praxn 1994:21:88 95. Cuyart GH, TovnsendM, Beman L3, Keller JL. A conparisonof likert and visualanaloeue{alesfof measurinschansein funcnon. ll. .lournalof Chroni€l Disease1987i40:1129

R/\n,,t

t

tr..

lt1 lta

9

Hanson L. BerglundM. Facros rnflucnc'ngtreatnent oulcode and patientsalisfadlonrnashorl lermpstchialicward A path analysn stud] of the mporranceolFricnt involvenentln lreahenl plannins. Elropeln archivcsof Psychialryand NeurologicalScience l9lt7j2l6126+7i H " n * o n L B e c l r n J V . L r J e n . r a r r (l . A r o e r . ' o r C . O h m d nR Patenr a tudesin short-ternpsychiatncc!re: Rellt1o.sro socral chnicll stnploms. and realment and psrchiatricback-qround, S.andinavica1985;72:193 l0l nodel. Acta Fsychiatrica oulcohe HoganTP.AwadAG. Subjeclneresponseloneuol€plicsand in shizophrenla A re exaninatro. companns tqo measures Mcdi.ine 1991223,1?5:. Psychologrcal Hoga. TP. Awad AG. Eastsood MR Early subjectiveEsponseand predrct,onof oulcomelo.eurolepticdrugtherap!ln s.htophren'a CanadianJournalof Pslchiatryl985il0:24G8. Jahn T. Mu$eay L Die slalistishe Ko.rrolle mdglicher Mediexpcrinenlalpsl-chologrschenSch kanentenernniiss in izophrcniestudien: Ein Vonchlag zur Ber€chrungvon Chlo. promaziniquivllenten zcilscnrift fiir Klim$he Pslchologic -rheory of Staiislics.vol. 2: Kendall MG, Stuart A. The Advanced London Gnfin- l97l Inlerenceand Relationship. Larsen.D. Aukisson,C, Hlrgreales.w. N guyen.T (l979) Ase$ffent of clienrrparienr satnfaclion:Delelopfrentoia gcncralscaleEvalualion and Progran Planning,l. 197 107 Lavik Nl Utilisationof nenral health servlccsovei a crvenpenod 1983:67:,104 11. Acta PstchiatncaScandinavica a$esnent Lavois M, NguyenT,AttkissonC Artituct in salisfaction Expenene in connunilt mennl healthetlings. Evaluarionand ProsramPllnning l9ltli4:ll9 50. Luiia RE The ralidily and reliabrhtyof the v$ual AnalosueMood S.ale tounal oiPsychialricRcsarch l9l5:t2:51 L Mclnryre K, Farcll M. Dalid A In-Patenl pslchiatrlccare The parienh liew. BntishJoumalolMedical lsy.holos) 1989i62:249 55 Ovemll JE. Corhan DR The brief psrrhiatricratine sclie Psycho logicalReportsl962il0:7998l:. PriebeS. Die Bedeulu.gder Paienl.nhci.uns Initiale Beserlu.C d ' o \ p r r d L rr * , h d ' r h . h e r l \ e r " o e C r r ' r n e . n H o g - e f el a o : . lriebe S. Cruyte$ T. The rol€ of the belpinEalliancein psychllric comnnnit! cxre A prospectivestudy Joumal of Nenous and MenlaD l i s e a sle9 9 l i l 8 l : 5 5 2L Piebe S, Cruyter! T. Paienlt rnd carcgiveGnririala$essnentsol dar nospilaltreahent and courseoi stnplons Comprehensrre Psychiatry1994i35:21+8. PriebeS. GruyremT Palienlj a$e$mentoftEahe.t prcdicrinrhos pitalzlton. Schizophrenia Bullelinl995jl 87 94 Sinen MM. Dysphoncreslonse1o neutolepl'ctfeatnenl rn schrzo phreni! and its pros.oslic signilicanceDseasesof tne NeNous SlsEn l9t6rl7l916 SteinhartI. PriebeS. Predicrionofhospnalizitionwilhin a pslchiatric communrrycare sysrm a 6ve lear study sdial Pstchiatryand Psychialic Epideniology 1992;27:27Gr. TansellaM, Mieiolo R. BalestrieriM. G.vloli I High and long-tern usersof the nentd heafthservic€sSocialPsychiatryl986i2l:96d t0l V.nPutlenT, Mly PRA Subj*riverespors asa predrctorol ourcome in ph.rnacothehp) Arcbiles olGeneral Psychialry1978i35:4?7 80 Var ?uuen T. May PRA, Marder SR, Withann LA Suhlectivc rcsponse10 antipslchoticdrugs Archirts of GeneralPsychialry to antipsychoticncdiVan lutienT. Mdy PRA. Mardcr SR.Response clrion: The dociois andthe consuneis liew AnericanJou.nalof Psychiatry1984jl4l: I G9 wi.glK. CooperJE.SarbnusN The MeasurenenlandCla$incatron of Psychrainc Syntroms.Canbridse CanbndgeUnilesnv Press.